What do large-scale patterns of extratropical atmospheric variability

imply about memory and predictability?

David W. J. ThompsonDepartment of Atmospheric Science

Colorado State University

What are the dominant “modes” of extratropical atmospheric variability?• The Pacific-North America pattern• The Northern Hemisphere annular mode (aka, the North Atlantic Oscillation or Arctic Oscillation)• The Southern Hemisphere annular mode

Why is it important to predict these modes?

To what extent does the midlatitude ocean impact these modes?

What alternative processes may drive low-frequency variability in these modes?

500-hPa height regressed on an index of the PNA

The Pacific-North America pattern

e.g., Wallace and Gutzler 1981

The Pacific-North America pattern

• Resembles a wave train emanating from the North Pacific.

• Excited by barotropic instability in the vicinity of the Pacific jet.

• Reminiscent of the extratropical atmospheric response to ENSO.

• Time series resembles a red-noise process with an e-folding timescale of ~10 days.

The Northern Hemisphere annular mode (NAM)

Sea-level pressure regressed on an index of the NAM

e.g., Thompson and Wallace 2000

The Southern Hemisphere annular mode (SAM)

850-hPa height regressed on an index of the SAM

e.g., Kidson, Karoly, Trenberth, etc.

The annular modes

• Characterized by zonally symmetric fluctuations in the extratropical atmospheric circulation that extend from the surface to the stratosphere.

• Driven by interactions between transient eddies and the mean flow of the extratropical atmosphere.

• The NAM and the NAO are different interpretations of the same mode of variability.

• The time series of the NAM (and SAM) resembles a red-noise process with an e-folding timescale of ~10 days. But the NAM (and SAM) also exhibits increased power at lower frequencies…

Why do we care about the annular modes?

• Extensive climate impacts.

• Time series exhibit low-frequency variability that appears to exceed the timescale of tropospheric variability.

Climate impacts of the NAM

Surface temperature regressed on the NAM index

Provided courtesy of Todd Mitchell, UW

Climate impacts of the SAM

700-hPa winds and surface temperature regressed on the SAM index (longest vector is ~4 m/s).

Thompson and Solomon 2002

Recent climate trends and the annular modes

e.g., Hurrell 1995; Thompson et al. 2000; Thompson and Solomon 2002

Recent trends in SH 500-hPa Z (left; Dec-May 1979-1998) and NH SLP (right; Jan-March 1968-1997).

What are the dominant “modes” of extratropical atmospheric variability?• The Pacific-North America pattern• The Northern Hemisphere annular mode (aka, the North Atlantic Oscillation or Arctic Oscillation)• The Southern Hemisphere annular mode

Why is it important to predict these modes?

To what extent does the midlatitude ocean impact these modes?

What alternative processes may drive low-frequency variability in these modes?

To what extent does the midlatitude ocean impact these modes?

1) To what extent can the persistence of the midlatitude oceans be explained via stochastic atmospheric processes?

2) To what extent does the midlatitude ocean impact the midlatitude atmosphere?

From Frankignoul and Hasselmann (1977; FH)

Hc

TF

dt

Td

pρλ ′−′

=′

F′T ′

λρcp

H

SST anomaly

White noise atmospheric forcing

Fixed thermal damping parameter

Density of seawater

Heat capacity of seawater

Fixed ocean mixed layer depth

The null hypothesis for SST persistence

where

FH model yields an e-folding timescale for SSTs of ~3-5 months

The null hypothesis of SST persistence, modified to include reemergence

(from Deser et al., in press J. Climate)

The FH model, when extended to account for re-emergence, yields SST persistence on the order of years.

Sample results from the extended FH model(from Deser et al., in press J. Climate)

Observed SSTA

Observed heat content anomaly

SSTA, original FH model

Heat content anomaly, modified FH model

Heat content anomaly, modified FH model with λ =0 in summer

Discrepancies occur where the mixed layer depth is shallow, near coastlines.

Observed vs. modeled autocorrelation

(from Deser et al., in press J. Climate)

Shading denotes r>0.3.

Observed

Theory

To what extent does the midlatitude ocean impact these modes?

1) To what extent can the persistence of the midlatitude oceans be explained via stochastic atmospheric processes?

2) To what extent does the midlatitude ocean impact the midlatitude atmosphere?

“We can now say with confidence that the extratropical ocean does indeed influence the atmosphere outside the boundary layer, but that this influence is of modest amplitude compared to internal atmospheric variability”Kushnir et al. 2002

The basic effects of atmosphere/ocean thermal coupling on midlatitude variability

(Barsugli and Battisti 1998)

• The original formulation by FH was extended by Barsugli and Battisti to include thermal feedbacks between the ocean and the atmosphere

- Ta, To are the surface temperatures of the atmosphere and ocean.

- a-d are dimensionless parameters that incorporate surface fluxes, radiative damping (b also includes any ocean dynamical feedback).- corresponds to the ratio of the heat capacity of the ocean mixed layer and the troposphere (~40). - N is atmospheric white noise forcing.

The basic effects of atmosphere/ocean thermal coupling on midlatitude variability

(Barsugli and Battisti 1998)

• Thermal coupling between the atmosphere and ocean increases the variance at low frequencies, and the persistence in both media.

• Simulations run with specified SST anomalies (i.e., without damped thermal coupling) yield spuriously large surface fluxes between the ocean and atmosphere.

Interpretation of AMIP-style simulations

Figure fromRodwell et al. 1999

Interpretation of AMIP-style simulations

Bretherton and Battisti (2000):

• The correlation between the simulated and observed NAO reflects the ensemble averaging, which acts to filter the noise in each ensemble member.

• As per the null hypothesis outlined in Barsugli and Battisti (1998), at low frequencies the correlation between the simulated and observed NAO approaches r=1.0 as the ensemble size approaches infinity.

• The results do not PROVE any predictability of the NAO beyond the persistence of the SST anomalies.

• Caveat: Barsugli and Battisti do NOT include advection by the gyre and/or thermohaline circulations (Czaja and Marshall 2000; Marshall et al. 2001).

What are the dominant “modes” of extratropical atmospheric variability?• The Pacific-North America pattern• The Northern Hemisphere annular mode (aka, the North Atlantic Oscillation or Arctic Oscillation)• The Southern Hemisphere annular mode

Why is it important to predict these modes?

To what extent does the midlatitude ocean impact these modes?

What alternative processes may drive low-frequency variability in these modes?

Stratosphere/troposphere coupling and the NAM

(from Baldwin and Dunkerton)

Concluding remarks

There is no conclusive evidence that variability in the midlatitude ocean has a significant impact on the

dominant modes of extratropical atmospheric variability.

Concluding remarks

The most compelling observational evidence of predictability on timescales longer than the limits of

deterministic weather prediction derives from stratosphere/troposphere coupling, not via coupling with

the midlatitude ocean.

Concluding remarks

1) A large fraction of the persistence of midlatitude SSTs is predicted by a simple model driven by stochastic atmospheric forcing.

caveat:Ocean dynamics likely plays an important role (e.g., the “inter-gyre gyre”), particularly in the vicinity of the western boundary currents.

2) There is increasing consensus that the atmospheric response to midlatitude SSTs is modest.

caveat:The annular modes have a pronounced impact on the climate of their respective hemispheres. A “modest” increase in the predictability of the NAM or SAM would likely be very useful.